Cloud systems can manage workloads by monitoring physical server and virtual machine utilization, including, for example, processor utilization, memory utilization, and storage utilization. If servers in a virtual machine cluster have low utilization, a cloud orchestration and management system might re-balance workloads by moving applications from a cluster with high utilization. Likewise, if servers in a virtual machine cluster have high utilization, the cloud orchestration and management system might re-balance workloads by moving the applications to a cluster with low utilization. The cloud orchestration and management system may add under-utilized clusters to highly utilized clusters in order to increase available capacity. Current cloud orchestration and management systems, however, cannot manage workloads for some types of services, including, for example, services that are delivered in real-time or near real-time.